The Superior Response and High Reproducibility of the Memristor-Integrated Low-Power Transparent SnO₂ Gas Sensor

Abstract

We present a SnO₂ gas sensor with an HfO₂ layer that exhibits enhanced performance and reliability for gasistor applications, combining a gas sensor and a memristor. The transparent SnO₂ gasistor with a 30 nm HfO₂ layer demonstrated low forming voltages (7.1 V) and a high response rate of 81.28% to 50 ppm of NO₂ gas, representing an approximately 174.86% increase compared to the response of 29.58% from the SnO₂ gas sensor without the HfO₂ layer. The device also showed improved power efficiency and exceptional long-term stability, with reproducibility tests over 10 days at 10 ppm NO₂ showing a minimal variation of 2.4%. These results indicate that the proposed transparent memristor with the 30 nm HfO₂ layer significantly enhances the device’s reliability and effectiveness for gasistor applications.

1. Introduction

The increasing interest in health monitoring has spurred the development of numerous transparent and wearable devices, including smart devices. Among these advancements, the ability to monitor the concentration of hazardous gases such as NO₂, CO, C₂H₆, etc., is crucial for real-time lung health assessment and disease prevention, as shown in Figure 1a,b. In particular, nitric dioxide (NO₂) gas sensors have gained considerable attention in the field of medical diagnostics and health monitoring [1,2,3]. The main interest in this area is diagnosing and tracking respiratory diseases such as lung disease, asthma, chronic obstructive pulmonary disease (COPD), and cystic fibrosis [4,5]. In individuals with asthma, when concentrations fall below normal levels, primary ciliary dyskinesia or cystic fibrosis may develop [6,7]. On the other hand, concentrations above normal levels may indicate the need for inhaled corticosteroids or the presence of viral infections. For these reasons, the demand for high-density, cost-effective, selective, and fast-acting gas sensors that can detect trace NO₂ gas in the air to protect human health and the environment is growing significantly [8,9,10,11]. For instance, Yeom et al. used indium oxide (In₂O₃) as a sensing layer to detect NO gas [12]. Similarly, Suehiro et al. used a carbon nanotube gas sensor to detect NO₂ gas [13]. In particular, SnO₂ gas sensors have recently gained attention as NO₂ gas sensors due to their high sensitivity and stability [14,15,16,17]. Alam et al. demonstrated the synthesis and characterization of Cu-SnO₂ nanoparticles using ultrasonic spray pyrolysis, highlighting the sensitivity and stability of SnO₂ nanocomposites for gas-sensing applications [18]. Despite these advancements, the requirements for high operating power and low reproducibility remain issues. To address the limitations of conventional gas sensors, we propose a novel SnO₂-based gas sensor integrated with memristor technology, applicable to various gas-sensing applications, as shown in Figure 1c. This study introduces a transparent SnO₂ gasistor, integrating a memristor structure into a SnO₂ gas sensor for enhanced performance. Various insulating layers have been utilized in prior studies; however, this work emphasizes the unique benefits of HfO₂ for gas sensor applications. The HfO₂ layer in our structure provides superior insulating properties and high thermal stability, effectively suppressing oxygen vacancy migration and stabilizing the CF [19]. This stabilization further enhances the gas-sensing performance by leveraging Joule heating effects, which strengthen the interaction between NO₂ molecules and the SnO₂ sensing layer, distinguishing our research from existing SnO₂-based sensor studies [20]. In this structure, the HfO₂ layer acts as an insulating material, stabilizing the CF and enhancing the overall gas-sensing performance for NO₂ detection. As shown in Figure 1d, the introduction of the HfO₂ layer leads to a more controlled and stable current reduction when exposed to NO₂ gas, resulting in improved performance, including gas response characteristics. The SnO₂ layer serves as both the top electrode (TE) and the active gas-sensing material, providing high sensitivity to NO₂. The inclusion of the HfO₂ layer significantly stabilizes the gas response while maintaining transparency and low power consumption. Our research explained the operating mechanism of this system, showing how the addition of the HfO₂ layer boosts the gas-sensing performance. This approach also creates opportunities for expanding the range of potential applications for different target gases, particularly in environmental and health-monitoring systems.

2. Materials and Methods

2.1. Fabrication of SnO₂ Gas Sensor with Memristor

Quartz substrates (ITASCO, 16 × 16 × 1 mm³) were cleaned by dipping them sequentially in acetone (Sigma-Aldrich, Saint Louis, MO, USA, 99.5%), methanol (Sigma-Aldrich, 99.8%), and deionized water solutions at 60 °C for 10 min each to remove residual impurities. After cleaning, a 100 nm-thick indium tin oxide (ITO) layer was deposited on a quartz substrate using RF magnetron sputtering (Korea Vacuum Tech., Gimpo, Republic of Korea, KVS-2000L) with an ITO target (VTEX, 99.99%) at a power of 100 W. The chamber conditions were maintained at 20 standard cubic centimeters per minute (sccm) of Argon (Ar) flow and 5 millitorrs (mTorr) of pressure. Subsequently, HfO₂ films with thicknesses of 20, 30, and 40 nm were deposited using RF magnetron sputtering with a HfO₂ target (ITASCO, 99.99%). Finally, a 100 nm SnO₂ layer was deposited at room temperature, maintaining the same Ar gas flow rate and pressure as in previous steps. The optical transmittance of the SnO₂ gasistor with 20, 30, and 40 nm HfO₂ layers was obtained using a UV–vis spectrophotometer (Cary 5000 UV–vis–NIR spectrophotometer, Agilent Technologies Inc., Santa Clara, CA, USA) with a wavelength accuracy of ±0.1 nm in a spectral range from 300 to 1100 nm.

2.2. Electrical and Gas Sensing Characterization of SnO₂ Gas Sensor with Memristor

The electrical properties were evaluated using a Pulse Generator Unit (Keithley, Cleveland, OH, USA, Keithley 4220) and a Semiconductor Characterization System (Keithley, Cleveland, OH, USA, Keithley 4200). The SnO₂ was connected to the output terminals of the Keithley 4200 SCS (current: ±0.025%, voltage: ±0.012%) via a triaxial cable using a tailored RF-compatible micro-positioner (MS Tech, Seoul, Republic of Korea, PB50). The bottom electrode (BE) was grounded. The device was placed in a customized aluminum chamber with a volume of approximately 30 cm³ to examine the response characteristics of the proposed devices to NO₂ gas in a controlled setting devoid of any external factors. The chamber was connected to a mass flow controller (MFC; DFPC1000, Daejeon City, Republic of Korea), which in turn was connected to tanks containing NO₂ and high-purity air (99.99%) gases. The NO₂ gas tank used nitrogen (N₂) as an inert gas, with a maximum allowable concentration of 50 parts per million (ppm). Prior to assessing the response characteristics, the humidity in the chamber was eliminated using high-purity air and heating. The temperature and humidity were maintained at a constant level of 25 °C and 20% relative humidity (RH). Before introducing NO₂ gas, high-purity air was injected into the chamber for 50 s to stabilize it. The NO₂ concentration was varied to 10, 20, 30, 40, and 50 ppm to examine its influence on the current, while the gas flow rate was set to 500 sccm.

3. Results and Discussion

Figure 2a illustrates the schematic structure of the proposed transparent SnO₂ gasistor. To achieve transparency, ITO was strategically selected as the bottom electrode (BE) due to its excellent optical transmittance and electrical conductivity, making it ideal for transparent electronic applications [21]. The SnO₂ layer served a dual function, acting as both the top electrode (TE) and the gas-sensing element.

First, to evaluate the transparency of the proposed structures, as illustrated in Figure 2b, the transmittance was measured in the wavelength range of 200 to 1100 nm. As a result, the transmittance in the visible range of 400–700 nm was 85.3%, 85.8%, and 85.9% for HfO₂ layer thicknesses of 20 nm, 30 nm, and 40 nm, respectively, indicating that the SnO₂ gasistor is suitable for transparent gas sensor applications [22]. The FE-SEM cross-sectional images in Figure 2c,d provide insight into the structural differences. Figure 2c shows the SnO₂ gas sensor without the HfO₂ layer, while Figure 2d depicts the SnO₂ gasistor with a 30 nm HfO₂ layer.

The electrical characteristics of the proposed SnO₂ gasistor were thoroughly examined using a Keithley 4200 parameter analyzer (Keithley, Cleveland, OH, USA, Keithley 4200) to assess its performance under varying conditions. Since the SnO₂ gasistor operates in a partially broken state of conductive filaments (CFs), the first step was to investigate the formation characteristics of the CFs at different applied voltages. As illustrated in Figure 3a, a forming process was conducted to activate the CFs within the HfO₂ layer, transitioning the device’s resistance from a high-resistance state (HRS) to a low-resistance state (LRS). This transition is critical for ensuring the proper function of the gasistor, as the CFs are responsible for modulating resistance during gas sensing operations. The forming voltage, required to induce this transition, showed a clear dependence on the thickness of the HfO₂ layer. Specifically, as the HfO₂ layer thickness increased, the forming voltage also increased. The measured forming voltages for HfO₂ layers with thicknesses of 20, 30, and 40 nm were 6.1, 7.1, and 8.2 V, respectively, demonstrating a positive correlation between layer thickness and the required forming voltage. This is due to the thicker HfO₂ layer increasing the migration distance of oxygen vacancies, necessitating a higher voltage for conductive filament formation. The energy required for oxygen ion movement through the thick insulating layer is higher, thereby enhancing the dielectric properties and resulting in an increased forming voltage [23]. Further investigation into the reset and set characteristics of the device revealed additional insights into its operational behavior. As shown in Figure 3b–d, the reset voltages for the 20, 30, and 40 nm HfO₂ layers were −1.74 V, −1.08 V, and −3.2 V, respectively, following the forming application. Similarly, the set voltages for these layers were recorded as 1.71 V, 0.96 V, and 2.12 V. Notably, the device with the 30 nm HfO₂ layer exhibited enhanced structural stability, reflected in the lower required reset and set voltages. This improvement in performance can be attributed to the optimal balance between the filament’s structural stability and the thickness of the HfO₂ layer, which facilitates more efficient CF formation and disruption. These findings highlight the importance of selecting the appropriate HfO₂ thickness to achieve optimal device performance, particularly in applications where energy efficiency and stability are critical.

The proposed device operates as a gas sensor in the HRS where the CF is broken. However, previous studies have shown that the resistance state of devices can be converted to an LRS even without gas injection. Thus, we first evaluated the stability of the broken CFs in the absence of gas concerning HfO₂ thickness. The current was monitored for 4000 s at 0.3 V without gas injection. Figure 4a shows the current value 0.3 V was applied for 4000 s. It was observed that the current remained stable for 4000 s when the HfO₂ thickness was 30 and 40 nm; however, when the thickness was reduced to 20 nm, the current became unstable. This instability can be attributed to the fact that as the thickness of the HfO₂ layer decreases, the insulating properties of the material become less effective, leading to increased leakage currents [24,25]. The transient response for each resistance state depending on the HfO₂ thickness is shown in Figure 4b–e. Gas sensing measurements were conducted in the HRS at RT, ensuring that the CF was in a broken state. This study focused on the NO₂ sensing characteristics across concentrations ranging from 10 to 50 ppm, chosen to align with standard levels frequently used in gasistor research, thereby facilitating direct comparisons with prior work [26]. At HfO₂ thicknesses of 20, 30, and 40 nm, a consistent increase in device response was observed as the NO₂ concentration increased. This proportional relationship suggests that higher gas concentrations enhance the interaction between gas molecules and the sensing material, thereby improving the sensing performance. This proportional relationship indicates that as the gas concentration increases, the interaction between gas molecules and the sensing material is enhanced, leading to improved sensing performance. This phenomenon is attributed to the formation of a Schottky barrier at the SnO₂-NO₂ interface, which results in an increase in the overall barrier height [27,28,29]. To accurately compare the changes in electrical conductivity concerning the applied voltage, the response (R) was calculated using the following formula:

$$\mathrm{R}={\displaystyle \frac{\left|{\mathrm{R}}_{\mathrm{a}\mathrm{i}\mathrm{r}}-{\mathrm{R}}_{\mathrm{g}\mathrm{a}\mathrm{s}}\right|}{{\mathrm{R}}_{\mathrm{a}\mathrm{i}\mathrm{r}}}}\times 100$$

(1)

As depicted in Figure 4f, the SnO₂ gasistor with a 30 nm HfO₂ layer demonstrated the highest response characteristics compared to the 20 nm and 40 nm layers. Notably, the SnO₂ gasistor with a 30 nm HfO₂ layer demonstrated a response of 81.28% at 50 ppm, compared to 39.77 and 43.81% for the 20 nm and 40 nm devices, respectively. In particular, the SnO₂ gasistor with a 30 nm HfO₂ layer demonstrated a 174.86% improvement in response compared to the 29.58% response of the SnO₂ gas sensor without the HfO₂ layer, underscoring the superior performance of the 30 nm HfO₂ layer. This phenomenon can be explained by the optimized thickness of the HfO₂ layer. In the HRS, the breaking of the CF generates localized heat through the Joule heating effect, which is efficiently transferred to the SnO₂ sensing layer, enhancing the interaction with NO₂ molecules [20,26]. The highest gas response observed at a thickness of 30 nm indicates that this thickness provides the best balance between heat generation and insulation properties. The optimized 30 nm thickness maximizes the interaction between the SnO₂ layer and NO₂ molecules, promoting stronger adsorption and significantly enhancing sensor performance

To assess the long-term stability of the SnO₂ gasistor without the HfO₂ layer and with a 30 nm HfO₂ layer, reproducibility measurements were carried out at 25 °C and 20% relative humidity in a sealed chamber, using an applied voltage of 0.3 V and a 10 ppm NO₂ gas concentration. In comparison to other studies, where gas sensing stability is typically evaluated over a period of 5 to 10 days, our study monitored the transient response of the device over 10 days. As shown in Figure 5a,b, the SnO₂ gas sensor without the HfO₂ layer exhibited a high variability of 22.7%. However, as shown in Figure 5c,d, the SnO₂ gasistor with the 30 nm HfO₂ layer demonstrated only 2.4% variability in response characteristics, indicating exceptional long-term stability. This low variability strongly suggests the high reproducibility of the gasistor, which is crucial for practical gas sensing applications. The stability is attributed to the effective retention of the CF within the HfO₂ layer, essential for maintaining reliable gas-sensing performance [29]. The consistent retention of CFs ensures stable operation, preventing degradation of the sensing element and enabling sustained, accurate NO₂ detection over prolonged periods. This high reproducibility and stability make the SnO₂ gasistor with the 30 nm HfO₂ layer a promising candidate for long-term gas sensing in real-environment applications, especially where continuous monitoring is required.

4. Discussion

The integration of an HfO₂ layer into the SnO₂ gas sensor plays a pivotal role in enhancing its gas-sensing performance and overall stability. The HfO₂ layer helps stabilize the CF that forms during resistive switching, allowing the sensor to reliably detect low concentrations of NO₂. By improving the modulation of resistance states, the HfO₂ layer significantly increases both the sensitivity and reproducibility. Notably, as shown in

Table 1. Performance of SnO₂ gasistor with HfO₂ layer of varying thicknesses.

HfO2 Thickness (nm)	Forming Voltage (V)	Reset Voltage (V)	Set Voltage (V)	Response at 10 ppm (%)
0	-	-	-	12.96
20	6.1	−1.74	1.71	18.17
30	7.1	−1.08	0.96	33.39
40	8.2	−3.2	2.12	28.73

, the SnO₂ gasistor with a 30 nm HfO₂ layer exhibits the highest performance, achieving a response of 33.39% at 10 ppm, which is considerably superior compared to the 18.17% and 28.73% responses observed with 20 nm and 40 nm layers, respectively. The proposed device demonstrated a recovery time of 87 s for 10 ppm NO₂ gas, addressing the slow recovery times reported in conventional SnO₂ gas sensors [30]. Additionally, long-term stability measurements over 10 days revealed only 2.4% variability, indicating significantly higher reproducibility compared to previous studies [31]. These improvements are crucial for ensuring that the sensor operates stably over extended periods, particularly in environments where conditions may fluctuate. This reliability is vital for applications requiring accurate gas detection in real-environmental conditions. Moreover, the incorporation of the HfO₂ layer reduces power consumption by facilitating efficient CF formation and retention, thereby minimizing energy requirements without compromising performance. The enhanced charge dynamics at the SnO₂-NO₂ interface, supported by the HfO₂ layer, contribute to improved sensing accuracy, especially at lower NO₂ concentrations. This low-power, high-sensitivity operation makes the SnO₂ gas sensor with an HfO₂ layer particularly suitable for transparent, energy-efficient gas-sensing applications, such as environmental monitoring, healthcare diagnostics, and industrial safety systems. In summary, the combination of the SnO₂ gas sensor and the HfO₂ layer offers a robust solution for long-term gas sensing. It improves response characteristics, enhances stability, and increases energy efficiency, meeting the growing demands for transparent, low-power electronics and continuous monitoring technologies. This design approach paves the way for advanced applications in environmental and industrial monitoring, providing enhanced performance without sacrificing energy efficiency.

5. Conclusions

In this study, we developed a high-performance SnO₂-based gasistor integrated with a transparent memristor structure, incorporating an HfO₂ layer for NO₂ gas sensing applications. The device demonstrated excellent gas-sensing capabilities, particularly with a 30 nm HfO₂ layer, which significantly improved the response and stability compared to the SnO₂ gas sensor without the HfO₂ layer. The gasistor with a 30 nm HfO₂ layer exhibited a response of 81.28% at 50 ppm of NO₂ gas, reflecting a 174.86% increase in sensitivity compared to the sensor without the HfO₂ layer. This improvement is primarily due to the stabilization of CFs within the HfO₂ layer, ensuring more reliable gas detection and enhanced long-term stability. The device maintained a stable current state for approximately 10,000 s, indicating its potential for extended gas sensing applications. Additionally, the memristor properties of the device amplified charge dynamics at the SnO₂-NO₂ interface, modulating the Schottky barrier and allowing the detection of lower NO₂ concentrations. The SnO₂ gasistor with a 30 nm HfO₂ layer also provided an optimal balance between forming voltage and response characteristics, offering superior performance. This approach presents new opportunities for transparent, reliable gas sensors in real-time environmental and health monitoring systems.